[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: !ERR and MPFIT

Pavel Romashkin <promashkin@cmdl.noaa.gov> writes:
> How about putting something like a conditional
> message, 'Number of iterations exceeded limits'
> in the user procedure, to signal MPFIT. Put no CATCH in user
> procedure. Then, in MPFIT you have CATCH statment followed by a
> graceful return. If a parameter is so out of bounds that user
> procedure signals, its unlikely you can obtain anything meaningful
> from MPFIT, then why not let it quit with CATCH? No common blocks,
> pointers or anything else. Fully self-contained, no conflicts due to
> multiple instances running, etc.

Good suggestion, and that actually what happens right now :-)

But I also want something a little more formal.  I still feel a little
attached to the common block implementation.  Still purely optional on
the part of the user.  We'll see multi-threaded IDL coming from a mile
away, if it ever comes.  I'll deal with it then. 

Right now I multiprocess MPFIT on my dual-CPU machine by running two
IDL sessions.  Works great!


Craig B. Markwardt, Ph.D.         EMAIL:    craigmnet@cow.physics.wisc.edu
Astrophysics, IDL, Finance, Derivatives | Remove "net" for better response